Changeset c9136d9


Ignore:
Timestamp:
Apr 26, 2022, 8:33:15 AM (2 years ago)
Author:
Peter A. Buhr <pabuhr@…>
Branches:
ADT, ast-experimental, master, pthread-emulation, qualifiedEnum
Children:
1b2adec, 57af3f3
Parents:
cd1a5e8
Message:

first proofread of chapter performance

File:
1 edited

Legend:

Unmodified
Added
Removed

  • doc/theses/mubeen_zulfiqar_MMath/performance.tex

    rcd1a5e8 rc9136d9  
    22\label{c:Performance}
    33
     4This chapter uses the micro-benchmarks from \VRef[Chapter]{s:Benchmarks} to test a number of current memory allocators, including llheap.
     5The goal is to see if llheap is competitive with the current best memory allocators.
     6
     7
    48\section{Machine Specification}
    59
    6 The performance experiments were run on three different multicore systems to determine if there is consistency across platforms:
     10The performance experiments were run on two different multi-core architectures (x86 and ARM) to determine if there is consistency across platforms:
    711\begin{itemize}
    812\item
    9 {\bf Nasus} AMD EPYC 7662, 64-core socket $\times$ 2, 2.0 GHz, GCC version 9.3.0
     13\textbf{Nasus} AMD EPYC 7662, 64-core socket $\times$ 2, 2.0 GHz, GCC version 9.3.0
    1014\item
    11 {\bf Algol} Huawei ARM TaiShan 2280 V2 Kunpeng 920, 24-core socket $\times$ 4, 2.6 GHz, GCC version 9.4.0
     15\textbf{Algol} Huawei ARM TaiShan 2280 V2 Kunpeng 920, 24-core socket $\times$ 4, 2.6 GHz, GCC version 9.4.0
    1216\end{itemize}
    1317
    1418
    15 \section{Existing Memory Allocators}\label{sec:curAllocatorSec}
    16 With dynamic allocation being an important feature of C, there are many stand-alone memory allocators that have been designed for different purposes. For this thesis, we chose 7 of the most popular and widely used memory allocators.
     19\section{Existing Memory Allocators}
     20\label{sec:curAllocatorSec}
     21
     22With dynamic allocation being an important feature of C, there are many stand-alone memory allocators that have been designed for different purposes.
     23For this thesis, 7 of the most popular and widely used memory allocators were selected for comparison.
     24
     25\subsection{glibc}
     26glibc~\cite{glibc} is the default gcc thread-safe allocator.
     27\\
     28\textbf{Version:} Ubuntu GLIBC 2.31-0ubuntu9.7 2.31\\
     29\textbf{Configuration:} Compiled by Ubuntu 20.04.\\
     30\textbf{Compilation command:} N/A
    1731
    1832\subsection{dlmalloc}
    19 dlmalloc (FIX ME: cite allocator with download link) is a thread-safe allocator that is single threaded and single heap. dlmalloc maintains free-lists of different sizes to store freed dynamic memory. (FIX ME: cite wasik)
     33dlmalloc~\cite{dlmalloc} is a thread-safe allocator that is single threaded and single heap.
     34It maintains free-lists of different sizes to store freed dynamic memory.
    2035\\
     36\textbf{Version:} 2.8.6\\
     37\textbf{Configuration:} Compiled with preprocessor @USE_LOCKS@.\\
     38\textbf{Compilation command:} @gcc -g3 -O3 -Wall -Wextra -fno-builtin-malloc -fno-builtin-calloc@ @-fno-builtin-realloc -fno-builtin-free -fPIC -shared -DUSE_LOCKS -o libdlmalloc.so malloc-2.8.6.c@
     39
     40\subsection{hoard}
     41Hoard~\cite{hoard} is a thread-safe allocator that is multi-threaded and using a heap layer framework. It has per-thread heaps that have thread-local free-lists, and a global shared heap.
    2142\\
    22 {\bf Version:} 2.8.6\\
    23 {\bf Configuration:} Compiled with pre-processor USE\_LOCKS.\\
    24 {\bf Compilation command:}\\
    25 cc -g3 -O3 -Wall -Wextra -fno-builtin-malloc -fno-builtin-calloc -fno-builtin-realloc -fno-builtin-free -fPIC -shared -DUSE\_LOCKS -o libdlmalloc.so malloc-2.8.6.c
    26 
    27 \subsection{hoard}
    28 Hoard (FIX ME: cite allocator) is a thread-safe allocator that is multi-threaded and using a heap layer framework. It has per-thread heaps that have thread-local free-lists, and a global shared heap. (FIX ME: cite wasik)
     43\textbf{Version:} 3.13\\
     44\textbf{Configuration:} Compiled with hoard's default configurations and @Makefile@.\\
     45\textbf{Compilation command:} @make all@
     46
     47\subsection{jemalloc}
     48jemalloc~\cite{jemalloc} is a thread-safe allocator that uses multiple arenas. Each thread is assigned an arena. Each arena has chunks that contain contagious memory regions of same size. An arena has multiple chunks that contain regions of multiple sizes.
    2949\\
     50\textbf{Version:} 5.2.1\\
     51\textbf{Configuration:} Compiled with jemalloc's default configurations and @Makefile@.\\
     52\textbf{Compilation command:} @autogen.sh; configure; make; make install@
     53
     54\subsection{pt3malloc}
     55pt3malloc~\cite{pt3malloc} is a modification of dlmalloc.
     56It is a thread-safe multi-threaded memory allocator that uses multiple heaps. pt3malloc heap has similar design to dlmalloc's heap.
    3057\\
    31 {\bf Version:} 3.13\\
    32 {\bf Configuration:} Compiled with hoard's default configurations and Makefile.\\
    33 {\bf Compilation command:}\\
    34 make all
    35 
    36 \subsection{jemalloc}
    37 jemalloc (FIX ME: cite allocator) is a thread-safe allocator that uses multiple arenas. Each thread is assigned an arena. Each arena has chunks that contain contagious memory regions of same size. An arena has multiple chunks that contain regions of multiple sizes.
     58\textbf{Version:} 1.8\\
     59\textbf{Configuration:} Compiled with pt3malloc's @Makefile@ using option ``linux-shared''.\\
     60\textbf{Compilation command:} @make linux-shared@
     61
     62\subsection{rpmalloc}
     63rpmalloc~\cite{rpmalloc} is a thread-safe allocator that is multi-threaded and uses per-thread heap. Each heap has multiple size-classes and each size-class contains memory regions of the relevant size.
    3864\\
     65\textbf{Version:} 1.4.1\\
     66\textbf{Configuration:} Compiled with rpmalloc's default configurations and ninja build system.\\
     67\textbf{Compilation command:} @python3 configure.py; ninja@
     68
     69\subsection{tbb malloc}
     70tbb malloc~\cite{tbbmallocmail
     71} is a thread-safe allocator that is multi-threaded and uses private heap for each thread. Each private-heap has multiple bins of different sizes. Each bin contains free regions of the same size.
    3972\\
    40 {\bf Version:} 5.2.1\\
    41 {\bf Configuration:} Compiled with jemalloc's default configurations and Makefile.\\
    42 {\bf Compilation command:}\\
    43 ./autogen.sh\\
    44 ./configure\\
    45 make\\
    46 make install
    47 
    48 \subsection{pt3malloc}
    49 pt3malloc (FIX ME: cite allocator) is a modification of dlmalloc. It is a thread-safe multi-threaded memory allocator that uses multiple heaps. pt3malloc heap has similar design to dlmalloc's heap.
    50 \\
    51 \\
    52 {\bf Version:} 1.8\\
    53 {\bf Configuration:} Compiled with pt3malloc's Makefile using option "linux-shared".\\
    54 {\bf Compilation command:}\\
    55 make linux-shared
    56 
    57 \subsection{rpmalloc}
    58 rpmalloc (FIX ME: cite allocator) is a thread-safe allocator that is multi-threaded and uses per-thread heap. Each heap has multiple size-classes and each size-class contains memory regions of the relevant size.
    59 \\
    60 \\
    61 {\bf Version:} 1.4.1\\
    62 {\bf Configuration:} Compiled with rpmalloc's default configurations and ninja build system.\\
    63 {\bf Compilation command:}\\
    64 python3 configure.py\\
    65 ninja
    66 
    67 \subsection{tbb malloc}
    68 tbb malloc (FIX ME: cite allocator) is a thread-safe allocator that is multi-threaded and uses private heap for each thread. Each private-heap has multiple bins of different sizes. Each bin contains free regions of the same size.
    69 \\
    70 \\
    71 {\bf Version:} intel tbb 2020 update 2, tbb\_interface\_version == 11102\\
    72 {\bf Configuration:} Compiled with tbbmalloc's default configurations and Makefile.\\
    73 {\bf Compilation command:}\\
    74 make
    75 
    76 \section{Experiment Environment}
    77 We used our micro becnhmark suite (FIX ME: cite mbench) to evaluate these memory allocators \ref{sec:curAllocatorSec} and our own memory allocator uHeap \ref{sec:allocatorSec}.
    78 
    79 \section{Results}
    80 FIX ME: add experiment, knobs, graphs, description+analysis
     73\textbf{Version:} intel tbb 2020 update 2, tbb\_interface\_version == 11102\\
     74\textbf{Configuration:} Compiled with tbbmalloc's default configurations and @Makefile@.\\
     75\textbf{Compilation command:} @make@
     76
     77% \section{Experiment Environment}
     78% We used our micro benchmark suite (FIX ME: cite mbench) to evaluate these memory allocators \ref{sec:curAllocatorSec} and our own memory allocator uHeap \ref{sec:allocatorSec}.
     79
     80\section{Experiments}
     81% FIX ME: add experiment, knobs, graphs, description+analysis
    8182
    8283%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    … …  
    8485%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    8586
    86 \subsection{Churn Benchmark}
    87 
    88 Churn benchmark tested memory allocators for speed under intensive dynamic memory usage.
    89 
     87\subsection{Churn Micro-Benchmark}
     88
     89Churn tests allocators for speed under intensive dynamic memory usage (see \VRef{s:ChurnBenchmark}).
    9090This experiment was run with following configurations:
    91 
    92 -maxS            : 500
    93 
    94 -minS            : 50
    95 
    96 -stepS           : 50
    97 
    98 -distroS         : fisher
    99 
    100 -objN            : 100000
    101 
    102 -cSpots          : 16
    103 
    104 -threadN         : \{ 1, 2, 4, 8, 16 \} *
    105 
    106 * Each allocator was tested for its performance across different number of threads. Experiment was repeated for each allocator for 1, 2, 4, 8, and 16 threads by setting the configuration -threadN.
    107 
    108 Results are shown in figure \ref{fig:churn} for both algol and nasus.
    109 X-axis shows number of threads. Each allocator's performance for each thread is shown in different colors.
    110 Y-axis shows the total time experiment took to finish.
     91\begin{description}[itemsep=0pt,parsep=0pt]
     92\item[thread:]
     931, 2, 4, 8, 16
     94\item[spots:]
     9516
     96\item[obj:]
     97100,000
     98\item[max:]
     99500
     100\item[min:]
     10150
     102\item[step:]
     10350
     104\item[distro:]
     105fisher
     106\end{description}
     107
     108% -maxS          : 500
     109% -minS          : 50
     110% -stepS                 : 50
     111% -distroS       : fisher
     112% -objN          : 100000
     113% -cSpots                : 16
     114% -threadN       : 1, 2, 4, 8, 16
     115
     116\VRef[Figure]{fig:churn} shows the results for algol and nasus.
     117The X-axis shows the number of threads.
     118Each allocator's performance for each thread is shown in different colors.
     119The Y-axis shows the total experiment time.
    111120
    112121\begin{figure}
    … …  
    118127\end{figure}
    119128
     129All allocators did well in this micro-benchmark, except for dmalloc on the ARM,
     130
     131
    120132%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    121133%% THRASH
    … …  
    124136\subsection{Cache Thrash}
    125137
    126 Thrash benchmark tested memory allocators for active false sharing.
    127 
     138Thrash tests memory allocators for active false sharing (see \VRef{sec:benchThrashSec}).
    128139This experiment was run with following configurations:
    129 
    130 -cacheIt        : 1000
    131 
    132 -cacheRep       : 1000000
    133 
    134 -cacheObj       : 1
    135 
    136 -threadN        : \{ 1, 2, 4, 8, 16 \} *
    137 
    138 * Each allocator was tested for its performance across different number of threads. Experiment was repeated for each allocator for 1, 2, 4, 8, and 16 threads by setting the configuration -threadN.
     140\begin{description}[itemsep=0pt,parsep=0pt]
     141\item[thread:]
     1421, 2, 4, 8, 16
     143\item[iterations:]
     1441,000
     145\item[cacheRW:]
     1461,000,000
     147\item[size:]
     1481
     149\end{description}
     150
     151% * Each allocator was tested for its performance across different number of threads.
     152% Experiment was repeated for each allocator for 1, 2, 4, 8, and 16 threads by setting the configuration -threadN.
    139153
    140154Results are shown in figure \ref{fig:cacheThrash} for both algol and nasus.
    … …  
    155169
    156170\subsection{Cache Scratch}
    157 
    158 Scratch benchmark tested memory allocators for program induced allocator preserved passive false sharing.
    159 
     171\label{s:CacheScratch}
     172
     173Scratch tests memory allocators for program-induced allocator-preserved passive false-sharing.
    160174This experiment was run with following configurations:
    161 
    162 -cacheIt        : 1000
    163 
    164 -cacheRep       : 1000000
    165 
    166 -cacheObj       : 1
    167 
    168 -threadN        : \{ 1, 2, 4, 8, 16 \} *
    169 
    170 * Each allocator was tested for its performance across different number of threads. Experiment was repeated for each allocator for 1, 2, 4, 8, and 16 threads by setting the configuration -threadN.
     175\begin{description}[itemsep=0pt,parsep=0pt]
     176\item[threads:]
     1771, 2, 4, 8, 16
     178\item[iterations:]
     1791,000
     180\item[cacheRW:]
     1811,000,000
     182\item[size:]
     1831
     184\end{description}
     185
     186% * Each allocator was tested for its performance across different number of threads.
     187% Experiment was repeated for each allocator for 1, 2, 4, 8, and 16 threads by setting the configuration -threadN.
    171188
    172189Results are shown in figure \ref{fig:cacheScratch} for both algol and nasus.
    … …  
    186203%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    187204
    188 \subsection{Speed Benchmark}
    189 
    190 Speed benchmark tested memory allocators for runtime latency.
    191 
     205\subsection{Speed Micro-Benchmark}
     206
     207Speed testa memory allocators for runtime latency (see \VRef{s:SpeedMicroBenchmark}).
    192208This experiment was run with following configurations:
    193 
    194 -maxS    :  500
    195 
    196 -minS    :  50
    197 
    198 -stepS   :  50
    199 
    200 -distroS :  fisher
    201 
    202 -objN    :  1000000
    203 
    204 -threadN    : \{ 1, 2, 4, 8, 16 \} *
    205 
    206 * Each allocator was tested for its performance across different number of threads. Experiment was repeated for each allocator for 1, 2, 4, 8, and 16 threads by setting the configuration -threadN.
     209\begin{description}[itemsep=0pt,parsep=0pt]
     210\item[max:]
     211500
     212\item[min:]
     21350
     214\item[step:]
     21550
     216\item[distro:]
     217fisher
     218\item[objects:]
     2191,000,000
     220\item[workers:]
     2211, 2, 4, 8, 16
     222\end{description}
     223
     224% -maxS    :  500
     225% -minS    :  50
     226% -stepS   :  50
     227% -distroS :  fisher
     228% -objN    :  1000000
     229% -threadN    : \{ 1, 2, 4, 8, 16 \} *
     230
     231%* Each allocator was tested for its performance across different number of threads.
     232%Experiment was repeated for each allocator for 1, 2, 4, 8, and 16 threads by setting the configuration -threadN.
    207233
    208234Results for speed benchmark are shown in 12 figures, one figure for each chain of speed benchmark.
    … …  
    337363%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    338364
    339 \subsection{Memory Benchmark}
    340 
    341 Speed benchmark tested memory allocators for their memory footprint.
    342 
    343 This experiment was run with following two configurations for each allocator:\\
    344 
     365\subsection{Memory Micro-Benchmark}
     366
     367This experiment is run with following two configurations for each allocator.
     368The difference between the two configurations is the number of producers and consumers.
     369Configuration 1 has one producer and one consumer, and configuration 2 has 4 producers where each producer has 4 consumers.
     370
     371\noindent
    345372Configuration 1:
    346 
    347 -threadA :  1
    348 
    349 -threadF :  1
    350 
    351 -maxS    :  500
    352 
    353 -minS    :  50
    354 
    355 -stepS   :  50
    356 
    357 -distroS :  fisher
    358 
    359 -objN    :  100000
    360 
    361 -consumeS:  100000\\
    362 
     373\begin{description}[itemsep=0pt,parsep=0pt]
     374\item[producer (K):]
     3751
     376\item[consumer (M):]
     3771
     378\item[round:]
     379100,000
     380\item[max:]
     381500
     382\item[min:]
     38350
     384\item[step:]
     38550
     386\item[distro:]
     387fisher
     388\item[objects (N):]
     389100,000
     390\end{description}
     391
     392% -threadA :  1
     393% -threadF :  1
     394% -maxS    :  500
     395% -minS    :  50
     396% -stepS   :  50
     397% -distroS :  fisher
     398% -objN    :  100000
     399% -consumeS:  100000
     400
     401\noindent
    363402Configuration 2:
    364 
    365 -threadA :  4
    366 
    367 -threadF :  4
    368 
    369 -maxS    :  500
    370 
    371 -minS    :  50
    372 
    373 -stepS   :  50
    374 
    375 -distroS :  fisher
    376 
    377 -objN    :  100000
    378 
    379 -consumeS:  100000
    380 
    381 Difference between the two configurations is the number of producers and consumers.
    382 Configuration 1 has one producer and one consumer.
    383 While configuration 2 has four producers where each producer has four consumers.
    384 
    385 \begin{table}[h!]
     403\begin{description}[itemsep=0pt,parsep=0pt]
     404\item[producer (K):]
     4054
     406\item[consumer (M):]
     4074
     408\item[round:]
     409100,000
     410\item[max:]
     411500
     412\item[min:]
     41350
     414\item[step:]
     41550
     416\item[distro:]
     417fisher
     418\item[objects (N):]
     419100,000
     420\end{description}
     421
     422% -threadA :  4
     423% -threadF :  4
     424% -maxS    :  500
     425% -minS    :  50
     426% -stepS   :  50
     427% -distroS :  fisher
     428% -objN    :  100000
     429% -consumeS:  100000
     430
     431\begin{table}[b]
    386432\centering
    387433    \begin{tabular}{ |c|c|c| }
    … …  
    411457
    412458Results for memory benchmark are shown in 16 figures, two figures for each of the 8 allocators, one for each configuration.
    413 Table \ref{table:mem-benchmark-figs} shows the list of figures that contain memory benchmar results.
     459Table \ref{table:mem-benchmark-figs} shows the list of figures that contain memory benchmark results.
    414460
    415461Each figure has 2 graphs, one for each experiment environment.
    … …  
    426472* These statistics are gathered by monitoring the \textit{/proc/self/maps} file of the process in linux system.
    427473
    428 For each subgraph, x-axis shows the time during the program lifetime at which the datapoint was generated.
     474For each subgraph, x-axis shows the time during the program lifetime at which the data point was generated.
    429475Y-axis shows the memory usage in bytes.
    430476
    431 For the experiment, at a certain time in the prgram's life, the difference betweem the memory requested by the benchmark (\textit{current\_req\_mem(B)})
    432 and the memory that the process has recieved from system (\textit{heap}, \textit{mmap}) should be minimum.
     477For the experiment, at a certain time in the program's life, the difference between the memory requested by the benchmark (\textit{current\_req\_mem(B)})
     478and the memory that the process has received from system (\textit{heap}, \textit{mmap}) should be minimum.
    433479This difference is the memory overhead caused by the allocator and shows the level of fragmentation in the allocator.
    434480
    … …  
    438484    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-1-prod-1-cons-100-cfa} }
    439485    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-1-prod-1-cons-100-cfa} }
    440 \caption{Memory benchmaark results with 1 producer for cfa memory allocator}
     486\caption{Memory benchmark results with 1 producer for cfa memory allocator}
    441487\label{fig:mem-1-prod-1-cons-100-cfa}
    442488\end{figure}
    … …  
    447493    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-1-prod-1-cons-100-dl} }
    448494    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-1-prod-1-cons-100-dl} }
    449 \caption{Memory benchmaark results with 1 producer for dl memory allocator}
     495\caption{Memory benchmark results with 1 producer for dl memory allocator}
    450496\label{fig:mem-1-prod-1-cons-100-dl}
    451497\end{figure}
    … …  
    456502    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-1-prod-1-cons-100-glc} }
    457503    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-1-prod-1-cons-100-glc} }
    458 \caption{Memory benchmaark results with 1 producer for glibc memory allocator}
     504\caption{Memory benchmark results with 1 producer for glibc memory allocator}
    459505\label{fig:mem-1-prod-1-cons-100-glc}
    460506\end{figure}
    … …  
    465511    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-1-prod-1-cons-100-hrd} }
    466512    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-1-prod-1-cons-100-hrd} }
    467 \caption{Memory benchmaark results with 1 producer for hoard memory allocator}
     513\caption{Memory benchmark results with 1 producer for hoard memory allocator}
    468514\label{fig:mem-1-prod-1-cons-100-hrd}
    469515\end{figure}
    … …  
    474520    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-1-prod-1-cons-100-je} }
    475521    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-1-prod-1-cons-100-je} }
    476 \caption{Memory benchmaark results with 1 producer for je memory allocator}
     522\caption{Memory benchmark results with 1 producer for je memory allocator}
    477523\label{fig:mem-1-prod-1-cons-100-je}
    478524\end{figure}
    … …  
    483529    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-1-prod-1-cons-100-pt3} }
    484530    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-1-prod-1-cons-100-pt3} }
    485 \caption{Memory benchmaark results with 1 producer for pt3 memory allocator}
     531\caption{Memory benchmark results with 1 producer for pt3 memory allocator}
    486532\label{fig:mem-1-prod-1-cons-100-pt3}
    487533\end{figure}
    … …  
    492538    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-1-prod-1-cons-100-rp} }
    493539    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-1-prod-1-cons-100-rp} }
    494 \caption{Memory benchmaark results with 1 producer for rp memory allocator}
     540\caption{Memory benchmark results with 1 producer for rp memory allocator}
    495541\label{fig:mem-1-prod-1-cons-100-rp}
    496542\end{figure}
    … …  
    501547    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-1-prod-1-cons-100-tbb} }
    502548    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-1-prod-1-cons-100-tbb} }
    503 \caption{Memory benchmaark results with 1 producer for tbb memory allocator}
     549\caption{Memory benchmark results with 1 producer for tbb memory allocator}
    504550\label{fig:mem-1-prod-1-cons-100-tbb}
    505551\end{figure}
    … …  
    510556    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-4-prod-4-cons-100-cfa} }
    511557    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-4-prod-4-cons-100-cfa} }
    512 \caption{Memory benchmaark results with 4 producers for cfa memory allocator}
     558\caption{Memory benchmark results with 4 producers for cfa memory allocator}
    513559\label{fig:mem-4-prod-4-cons-100-cfa}
    514560\end{figure}
    … …  
    519565    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-4-prod-4-cons-100-dl} }
    520566    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-4-prod-4-cons-100-dl} }
    521 \caption{Memory benchmaark results with 4 producers for dl memory allocator}
     567\caption{Memory benchmark results with 4 producers for dl memory allocator}
    522568\label{fig:mem-4-prod-4-cons-100-dl}
    523569\end{figure}
    … …  
    528574    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-4-prod-4-cons-100-glc} }
    529575    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-4-prod-4-cons-100-glc} }
    530 \caption{Memory benchmaark results with 4 producers for glibc memory allocator}
     576\caption{Memory benchmark results with 4 producers for glibc memory allocator}
    531577\label{fig:mem-4-prod-4-cons-100-glc}
    532578\end{figure}
    … …  
    537583    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-4-prod-4-cons-100-hrd} }
    538584    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-4-prod-4-cons-100-hrd} }
    539 \caption{Memory benchmaark results with 4 producers for hoard memory allocator}
     585\caption{Memory benchmark results with 4 producers for hoard memory allocator}
    540586\label{fig:mem-4-prod-4-cons-100-hrd}
    541587\end{figure}
    … …  
    546592    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-4-prod-4-cons-100-je} }
    547593    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-4-prod-4-cons-100-je} }
    548 \caption{Memory benchmaark results with 4 producers for je memory allocator}
     594\caption{Memory benchmark results with 4 producers for je memory allocator}
    549595\label{fig:mem-4-prod-4-cons-100-je}
    550596\end{figure}
    … …  
    555601    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-4-prod-4-cons-100-pt3} }
    556602    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-4-prod-4-cons-100-pt3} }
    557 \caption{Memory benchmaark results with 4 producers for pt3 memory allocator}
     603\caption{Memory benchmark results with 4 producers for pt3 memory allocator}
    558604\label{fig:mem-4-prod-4-cons-100-pt3}
    559605\end{figure}
    … …  
    564610    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-4-prod-4-cons-100-rp} }
    565611    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-4-prod-4-cons-100-rp} }
    566 \caption{Memory benchmaark results with 4 producers for rp memory allocator}
     612\caption{Memory benchmark results with 4 producers for rp memory allocator}
    567613\label{fig:mem-4-prod-4-cons-100-rp}
    568614\end{figure}
    … …  
    573619    \subfigure[Algol]{ \includegraphics[width=0.9\textwidth]{evaluations/algol-perf-eps/mem-4-prod-4-cons-100-tbb} }
    574620    \subfigure[Nasus]{ \includegraphics[width=0.9\textwidth]{evaluations/nasus-perf-eps/mem-4-prod-4-cons-100-tbb} }
    575 \caption{Memory benchmaark results with 4 producers for tbb memory allocator}
     621\caption{Memory benchmark results with 4 producers for tbb memory allocator}
    576622\label{fig:mem-4-prod-4-cons-100-tbb}
    577623\end{figure}
Note: See TracChangeset for help on using the changeset viewer.